Ovid MEDLINE Instruction can be Evaluated Using a Validated Search Assessment Tool

Abstract

Objective – To determine the construct validity of a search assessment instrument that is used to evaluate search strategies in Ovid MEDLINE.

Design – Cross-sectional, cohort study.

Setting – The Academic Medical Center of the University of Michigan.

Subjects – All 22 first-year residents in the Department of Pediatrics in 2004 (cohort 1); 10 senior pediatric residents in 2005 (cohort 2); and 9 faculty members who taught evidence based medicine (EBM) and published on EBM topics.
Methods – Two methods were employed to determine whether the University of Michigan MEDLINE Search Assessment Instrument (UMMSA) could show differences between searchers’ construction of a MEDLINE search strategy.

The first method tested the search skills of all 22 incoming pediatrics residents (cohort 1) after they received MEDLINE training in 2004, and again upon graduation in 2007. Only 15 of these residents were tested upon graduation; seven were either no longer in the residency program, or had quickly left the institution after graduation. The search test asked study participants to read a clinical scenario, identify the search question in the scenario, and perform an Ovid MEDLINE search. Two librarians scored the blinded search strategies.

The second method compared the scores of the 22 residents with the scores of ten senior residents (cohort 2) and nine faculty volunteers. Unlike the first cohort, the ten senior residents had not received any MEDLINE training. The faculty members’ search strategies were used as the gold standard comparison for scoring the search skills of the two cohorts.

Main Results – The search strategy scores of the 22 first-year residents, who received training, improved from 2004 to 2007 (mean improvement: 51.7 to 78.7; t(14)=5.43, P<0.0001). The graduation scores were also significantly higher for this first cohort compared to the second cohort, who received no training (median 85.0 vs. 65.0; Wilcoxon chi-square(1)=4.09, P=0.043). The graduation scores of the first cohort were similar to those of faculty volunteers (Wilcoxon chi-square(1)=3.82, P=0.050) .

Conclusion – According to the authors, “the results of this study provide evidence for the validity of an instrument to evaluate MEDLINE search strategies” (p. 81), since the instrument under investigation was able to measure improvements and differences in the search performances of the study’s participants. A validated search assessment instrument can effectively measure improvements in residents’ search skills to demonstrate training effectiveness, as well as satisfy practice-based learning competency requirements from the Accreditation Council for Graduate Medical Education.